Edit

app.py

@@ -1,5 +1,4 @@
 import torch
-import gradio as gr
 from fastapi import FastAPI, File, UploadFile
 from fastapi.responses import JSONResponse
 from transformers import ConvNextForImageClassification, AutoImageProcessor
@@ -29,49 +28,28 @@ processor = AutoImageProcessor.from_pretrained("facebook/convnext-base-224")
 # FastAPI app
 app = FastAPI()
 
-# Helper function for processing the image
+# Prediction helper
 def predict(image: Image.Image):
-    # Preprocess the image
     inputs = processor(images=image, return_tensors="pt")
-    
-    # Perform inference
     with torch.no_grad():
         outputs = model(**inputs)
         predicted_class = torch.argmax(outputs.logits, dim=1).item()
-    
     return predicted_class, class_names[predicted_class]
 
-# FastAPI endpoint to handle image upload and prediction
+# Endpoint: /predict
 @app.post("/predict/")
 async def predict_endpoint(file: UploadFile = File(...)):
     try:
-        # Read and process the image
         img_bytes = await file.read()
-        img = Image.open(io.BytesIO(img_bytes))
-        
-        # Get the prediction
+        img = Image.open(io.BytesIO(img_bytes)).convert("RGB")
         predicted_class, predicted_name = predict(img)
-        
-        # Return the result as JSON
-        return JSONResponse(content={"predicted_class": predicted_class, "predicted_name": predicted_name})
-    
+        return JSONResponse(content={
+            "predicted_class": predicted_class,
+            "predicted_name": predicted_name
+        })
     except Exception as e:
         return JSONResponse(content={"error": str(e)}, status_code=500)
 
-# Gradio function to integrate with the FastAPI prediction
-def gradio_predict(image: Image.Image):
-    predicted_class, predicted_name = predict(image)
-    return f"Predicted Class: {predicted_name}"
-
-# Gradio Interface
-iface = gr.Interface(fn=gradio_predict, inputs=gr.Image(type="pil"), outputs=gr.Textbox())
-
-# Serve Gradio interface on FastAPI
-@app.get("/gradio/")
-async def gradio_interface():
-    return iface.launch(share=True, inline=True)
-
-# Run the FastAPI app using Uvicorn
-if __name__ == "__main__":
-    import uvicorn
-    uvicorn.run(app, host="0.0.0.0", port=8000)
+# Required for Hugging Face Spaces (do NOT run uvicorn manually)
+# Just expose the app
+app = app